Rolling Element Bearing Methodology

Application Guide
Bently Nevada Machinery Condition Monitoring

Objective
This document is intended to be the complete source for understanding Bently Nevada’s
methodology on monitoring/managing rolling element bearings. The intention is to reduce the
number of sources/locations where such information is stored.

Deliverables and CTQs

This document describes in detail Bently Nevada’s methodology for managing Rolling Element
Bearings which has been proven over the past years using Trendmaster Pro, 1900/65A, vbX/SCOUT
instruments along with Ascent software and System 1 Evolution. This methodology is accepted
throughout Bently Nevada and multiple industries.

Introduction
This document is intended to be the master document for Bently Nevada’s methodology for
management of rolling element bearings (REB’s). To avoid the repetitive reworking and releasing of
multiple documents, all other documents that require mention of Bently Nevada’s REB management
methodology should simply refer to this document rather than duplicating parts of it.
Bently Nevada’s REB management methodology includes the important aspects of transducer
types/usage, transducer mounting, monitor and software configuration and processing techniques. It is
based on both the principles of rotating machinery behavior and the vibration characteristics of rolling
element bearings. It combines traditional strategies for detecting rotor-related problems (such as
unbalance and misalignment) with specialized methodologies (such as enveloping or demodulation)
for identifying rolling-element bearing defects. The monitoring system must provide data for
detecting and preventing rotor and bearing related problems and give adequate advanced warning of
these problems so corrective action may be initiated. The system must be capable of discriminating
among bearing faults, and give an early warning when a lubrication problem is present. All this is
accomplished using an accelerometer and presenting data in acceleration, velocity and acceleration
enveloping (demodulation) units.
Note that in this document, both the American and European conventions for marking decimal places
are used. Numerical values may employ either the "." or "," mark, as in 1.41 or 1,41.

Document: 164934
Rev. C
Rolling Element Bearing Methodology
Application Guide

Copyright 2019   Baker Hughes, a GE company, LLC ("BHGE")

All rights reserved.
Bently Nevada, Orbit Logo, System 1, SCOUT, Ranger, vbSeries, 6Pack and Velomitor are registered
trademarks of BHGE in the United States and other countries. All product and company names are
trademarks of their respective holders. Use of the trademark does not imply any affiliation with or
endorsement by the respective holders.
The information contained in this document is the property of BHGE and its affiliates; and is subject to
change without prior notice. It is being supplied as a service to our customers and may not be altered
or its content repackaged without the express written consent of BHGE.
This product or associated products may be covered by one or more patents. See Bently.com/legal.

Contact Information
Mailing Address 1631 Bently Parkway South
Minden, Nevada USA  89423
Telephone 1.775.782.3611
1.800.227.5514 (US only)
Internet Bently.com

2/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

Contents
1.  Rolling Element Bearing Basics 4
1.1  Rolling Element Bearing (REB) Principles 4
1.2  Basic Bearing Geometry 5
1.3  Load, Stresses, and Bearing Life 6
1.4  Causes of Failure 8
1.5  Fatigue Failure in Rolling Element Bearings 10
1.6  Failure Stages 12
1.7  Bearing Defect (Fault) Frequencies 13
2.  Transducers Types and Usage 19
2.1  Transducer Location and Quantity 19
2.2  Sensor selection 21
2.3  Accelerometer Vibration Transducer Mounting 26
3.  Alarm Levels 29
4.  Vibration Settings for Bently Nevada Monitors in REB Applications 31
4.1  3500 Monitor 31
4.2  3701 Monitor 31
4.3  2300 Monitor 31
4.4  1900/65A Monitor 32
4.5  VbOnlinePro Monitor 32
4.6  Trendmaster Pro Dynamic Scanning Module (DSM) TIM-line 33
4.7  RangerPro (Wireless) 33
4.8  SCOUT Portable Data Collector (PDC) 34
5.  Monitoring Slow Speed Machines 35
6.  Conclusion 36
6.1  References 36

3/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

1.   Rolling Element Bearing Basics

1.1  Rolling Element Bearing (REB) Principles
All bearings serve the primary function of keeping a rotating part from contacting another part that
may be either rotating at a different speed or stationary. A bearing must provide stiffness to keep the
rotor in position, and it must do so while minimizing friction. Fluid-film bearings do this by providing
a thin layer of lubricating fluid between the rotor and journal. Rolling element bearings do this using a
set of balls or rollers (also called elements), lubricated by grease or oil, that keep the inner and outer
rings separated.

Rolling element bearings find many uses in today's machinery. They are used in applications where
rotation speed is too low to create a reliable fluid-film support, where the cost or weight of a fluid-
film bearing and its associated lubricant supply system cannot be justified, or in applications that
require close tolerances between stationary and rotating parts. They can be found in motors, wind
turbine gearboxes, gas turbines, pumps, and many other machines. Some of the reasons rolling
element bearings are used are: low starting friction, low operating friction, ability to support loads at
low (even zero) speed, simpler lubrication requirements, and the ability to support both radial and
axial loads in the same bearing. When some of these factors are important, rolling element bearings
may be in use. For example, industrial gas turbines use fluid-film bearings, whereas aircraft gas turbines
use rolling element bearings because of weight and stiffness constraints. Large critical machinery in
power plants have fluid-film bearings, while for cost reasons many smaller motors and other
supporting machines have rolling element bearings.
Rolling element bearings are usually subjected to both static and dynamic radial loads in combination
with axial or thrust loads. The radial loads are transmitted from the shaft, through the inner ring, shared
by several rolling elements, and transmitted to the outer ring and into the housing. The number of
rolling elements that share the load transmission will depend on the element spacing in the bearing
and on the tolerance class of the bearing. High quality, close tolerance bearings will have more rolling
elements involved simultaneously with supporting the load and load transmission than will a loose
tolerance bearing.
The region of the bearing that is involved in the load transmission is referred to as the load zone. The
load zone represents the area of the bearing involved in the transmission of radial loads. The angular

4/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide
size of the load zone depends on the precision of the bearing. High precision bearings have wider
load zones than low precision bearings. For the portion of time (rotation) that rolling elements are
outside the load zone, they do not transmit radial load.
Rolling element bearings come in special designs for radial loads, thrust loads, or a combination of
these loads. They differ in the shape and angle of contact of the elements, and in the geometry of the
races (the groove or surface in the rings that the elements ride on). Most simple ball bearings, while
primarily designed for radial loads, can handle a limited amount of thrust loading as well.
When more thrust must be accommodated, angular contact ball bearings are often used. These
bearings have an extended race design that provides more support for the balls in thrust loading.
Angular contact ball bearings can be installed in back-to-back pairs to provide thrust capability in both
directions.
High loads require more contact area between elements and races to keep stresses within allowable
limits. To accomplish this, cylindrical rollers are used instead of spherical balls. This geometry provides
a much larger contact area, and lower stresses for the same loading.
For combinations of high radial and high thrust loads, tapered roller bearings are used. For example,
tapered roller bearings are often used as automobile wheel bearings because of the need to handle
high radial loads (loads from braking and bumping) and high thrust loads (loads from cornering).
For very high thrust loads, special thrust bearings are used where the rings are separated axially instead
of radially. The thrust bearings can have balls, tapered rollers, or needles as rolling elements.
Most common rotating machine applications involve relatively simple radial or angular contact ball
bearings. These kinds of bearings are commonly used in industrial machines such as electric motors,
pumps, and fans. They are also the predominant type used in aeroderivative gas turbines.

1.2  Basic Bearing Geometry

Rolling element bearings come in a wide variety of shapes (geometries) that are optimized for
different applications. Ball bearings are used for moderate loads and produce nearly “point” contact
between the elements and races. This contact “point” is actually a small, elliptically-shaped contact
area. Cylindrical and tapered roller bearings are used for relatively high loads and produce a much
larger contact area. Some rollers are crowned or spherical and have shallow relative radii, producing an
intermediately sized contact area for better load distribution and higher load carrying capacity than a
simple ball bearing.
It is beyond the scope of this paper to explore the different geometries in detail. Consult a reference
such as Harris [Ref. #1] for more information. See "References" on page 36. Also, most bearing
manufacturers have websites containing a wealth of information.
A simple ball bearing serves to illustrate the geometry of a simple REB. Figure 1 shows a cross section
of a radial ball bearing that is subjected to a simultaneous radial and thrust load. The clearance and
axial displacement are exaggerated for clarity.

5/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

Figure 1 - 1:  Basic Geometry of a Ball Bearing

The clearance and axial displacement are exaggerated for clarity. Important parameters are the pitch
diameter, D, the contact angle, α (alpha), and the element diameter, ɗ (not shown). The cage serves to
control the spacing of the elements.
The bearing consists of an outer ring and race, an inner ring and race, cage and a set of rolling
elements. In most applications, the outer ring is installed with a slight interference fit in the machine
structure and does not rotate. The inner ring carries the rotating shaft. In some cases, the inner ring is
stationary and the outer ring rotates. In rare applications, both the outer and inner rings rotate at
different angular velocities. The cage serves to keep the rolling elements separated and evenly spaced
and, in roller bearings, to control the orientation of the rollers.
Four parameters are used to calculate characteristic bearing frequencies that are important for
diagnostic work . See "Bearing Defect (Fault) Frequencies" on page 13. The pitch diameter, D, is the
distance across the bearing between element centers. The contact angle, α (alpha), is a function of the
ratio of the radial load to the axial load and the curvature of the elements and race surfaces. For purely
radial loading and symmetric races, the contact angle is zero. The element diameter,ɗ, is the diameter
of the rolling element, and ր is the number of elements.

1.3  Load, Stresses, and Bearing Life

Properly designed and carefully lubricated fluid-film bearings can have virtually infinite life. However,
given enough time all REBs will eventually fail through a fatigue mechanism that can be greatly
accelerated by poor installation, overloading, improper lubrication, or contamination.
A load that originates in a rotor shaft is transmitted in a rolling element bearing from the inner ring
and race, through the rolling element, and through the outer race and ring into the surrounding
structure. Because of the geometry of the bearing parts, contact between the rolling elements and the
races occurs in extremely small physical dimensions. Theoretically, an infinitely stiff spherical ball will
contact an infinitely stiff race only at a point. Given any load at all, this would theoretically produce an
infinite stress.
In reality, mutual elastic deformation of the ball and race will produce an elliptically shaped zone of
contact over which the load is distributed. Similarly, a roller in contact with a race of exactly the same
length will produce a rectangular contact zone. The deformation in the “point” contact zone produces

6/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide
a compressive stress distribution that was first analyzed by Hertz in 1881. Because of this, these stresses
are referred to today as Hertzian contact stresses.
The maximum compressive stress occurs at the center of the contact zone and can be quite large in
modern bearing designs. Harris stated that rolling element bearings often operate continuously with
compressive stresses exceeding 200 ksi (1400 N/mm2)(Ref.#1). See "References" on page 36.The
microscopic surface roughness of the contacting elements further complicates the picture, as does the
presence of any kind of debris that may become trapped between the rolling element and the race.
It is, however, not the surface compressive stress that leads directly to failure. According to the
distortion energy theory, the effective (Von Mises) stress is responsible for material yielding.
Surprisingly, for the type of loading that occurs in rolling element bearings, the maximum effective
stress occurs at some depth beneath the surface of the contacting elements. (Ref #1) See "References"
on page 36.
As a rolling element passes over a location on a race, the material beneath the surface of the race is
subjected to a high effective stress for a short time during element passage. This cycling of stress
occurs once per element passage at the location. Simultaneously, a location on the rolling element
itself is also subjected to a cycle of subsurface stress twice per revolution of the element. As the stress
cycling occurs, the high subsurface stresses can produce subsurface micro-yielding and phase
transformations that favor the development of a fatigue crack. Because of the nature of the element
loading and the stresses that exist in rolling element bearings, they often fail due to the propagation
of fatigue cracks that originate below the surface.
The cleanliness of bearing lubricant can also have a strong influence on bearing life. Loewenthal has
shown that ultra-filtration can significantly prolong bearing life (Ref#2). See "References" on page 36.
Even micron-sized debris that becomes trapped between a rolling element and a race can cause local
plastic yielding and the formation of a small pit in either or both surfaces. The relatively small radius of
the pit boundary acts as a stress concentration factor and greatly increases the likelihood of initiation
of a fatigue crack.
Because of this complexity, rolling element-bearing failure is very much a statistical process. For any
individual bearing, the timing of fatigue crack origin can depend on the details of metallurgical
microstructure such as the location and size of inclusions, phase changes, and discontinuities. It can
also depend on deviations of surface geometry or surface smoothness.
Because of the complex and concentrated loading in the various parts of a rolling element bearing,
bearing lifetime is limited and is very sensitive to load. Typically, rolling element bearing life, L, can be
expressed as:

Figure 1 - 2:  Eguation #1
Where L R is the lifetime at the rated load, FR , and F is the actual load applied to the bearing. The
exponent of 3.33 applies to roller bearings, whereas an exponent of 3 replaces it for ball bearings. L R
is usually expressed in terms of a large number of revolutions, such as 90 million. (While this seems like
a large number, 90 million revolutions will be achieved by a 3600 rpm machine in a little over 17 days.)
See "References" on page 36.

7/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide
The lifetime, L, from the above equation is usually stated in terms of a reliability number such as 90%
or 50% for a group of apparently identical rolling element bearings operating under identical loads
and speeds. For example, the L10 (sometimes called B10) life is defined as the rated life where 10% of
the group will show evidence of fatigue failure and 90% of the group will show no evidence of failure
(Ref#3). See "References" on page 36. Rated lifetimes are based on bearing operation under ideal
conditions at steady load in clean environments with clean lubricating oils.
Note that from Equation #1 above, bearing life is related to about the 3rd power of the load. Thus, a
load increase of 50% above rated load results in a 70% reduction in expected lifetime. This
computation assumes a steady load on the bearing. Dynamic or shock loading of rolling element
bearings can drastically reduce bearing life.

1.4  Causes of Failure

Most real-world bearings do not operate under ideal conditions and they fail well before reaching
their theoretical design life. Final rolling element bearing failure is almost always a result of fatigue,
but premature failure can be related to a number of root causes. When analyzing premature rolling
element bearing failures, it is important not only to detect that the bearing is failing, but also to
determine the underlying cause of that failure. Most premature bearing failures can be attributed to
one or more of the following causes:

 l Excessive loading, either static or dynamic

 l Improper lubrication
 l External contamination
 l Improper installation
 l Incorrect bearing selection for application
 l Exposure to vibration while not rotating
 l Passage of electric current through the bearing

This list includes the major causes of premature bearing failure and can be used as an initial guide to
determine the reason for a bearing failure. To ensure success, elimination of premature bearing failures
must be a major goal of any predictive maintenance program.

Excessive Loading
As shown above, bearing life is very sensitive to loading. Bearing loads can be classified into static
loads, which have constant magnitude and direction, and dynamic loads, which can vary in both
magnitude and direction. To properly size a bearing at the design stage, a machine designer must
consider both types of loading and their influence on the predicted life of the bearing. If during
service the loads in a machine deviate significantly from design values, bearing life will be affected.
Misalignment is an important source of excessive static load. Properly designed and sized couplings
will accommodate misalignment to some degree, but if misalignment exceeds the capability of the
coupling, then excessive loads may appear at the bearing. Another important source of static load is a
belt drive.
Dynamic loads are introduced by many rotor malfunctions, but may also result from the work the
machine is doing. Unbalance, rub, or aerodynamic or fluid-induced instability can produce dynamic

8/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide
loads that exceed design limits and reduce bearing life. Also, coupling malfunctions may introduce
excessive static and dynamic loads.

Improper Lubrication
Rolling element bearings depend on an extremely thin layer of lubricant to keep rolling element/race
contact to a minimum. The rolling elements actually skid to a degree, and the lubricant forms a type of
elastohydrodynamic wedge that keeps the elements separated. The separation of elements is on the
order of a few times the typical surface roughness, less than a micron (or a few micro inches).
The lubricating fluid also serves the purpose of removing the heat that is generated by shearing in this
very small gap between elements and races. If the lubrication supply is insufficient, then metal-to-
metal contact will occur, creating wear and possible smearing of races. The lubricant must be clean and
of the proper type for the load and operating temperature. Rolling element bearings can survive for
long periods of time quite well with relatively small amounts of lubricant. Excessive lubrication can be
harmful by causing high temperatures from the lubricant shearing stresses and can actually lead to
bearing failure.

External Contamination
Cleanliness of lubricant is vital to long bearing life (Ref #2). See "References" on page 36.If dirt
particles get into the bearing either directly or through the oil system, then the particles can become
trapped between the rolling element and the race. If the particle has sufficient hardness, then the
extreme stress produced at the particle contact will cause local plastic deformation of the race and/or
element material. This will produce a permanent pit that will act as a potential crack nucleation site. It
is also possible for a large particle to jam in such a way as to reduce element rolling action and
increase sliding motion. During sliding motion, the dirt particle can act like a plow, creating a
permanent groove in the race surface.

Improper Installation
Many bearings have been doomed during installation. Bearings must be handled carefully and installed
in accordance with manufacturer’s specifications. If interference fits are used on the bearing rings, care
must be taken to avoid deformation or cocking of the rings during installation. Cocked or misaligned
bearing rings will produce abnormal loads in the bearing. Incorrect fits can also subject the bearing to
abnormal loads. And although this may seem obvious, bearings should never be pressed on in such a
way that the press load is delivered across the rolling elements. The high press loads can cause the
balls to permanently dimple (brinell) the races.

Incorrect Bearing for Application

If the designer fails to consider all the possible loads that can occur in service, the bearing selected for
the application could be incorrect. If the service loads exceed the design intention, then the bearing
will fail prematurely. It is important for the designer to be aware of the static and dynamic loading that
may be present in service.
On the other end of the load spectrum is the situation where the bearing is very lightly loaded. In this
condition the roller skids rather than rolls, which is also not desired.
It is also possible that during assembly or repair a wrong bearing was installed.

9/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

Exposure to Vibration While Not Rotating

Brinelling is a form of non-rotating load damage. When a rotor is stopped, the rolling elements in
bearings remain in one position. If the machine is subjected to a shock load, the elements (especially
balls) can plastically deform the races, creating a permanent set of pits. This has happened to new
machines being shipped by rail or truck. The machines arrive at the plant only to find that the bearings
are ruined. To prevent this kind of damage, vulnerable machinery is usually rotated on turning gear
during shipment.
It is normal that a spare (stand-by) machine is exposed to some level of vibration while not operating.
Its static load will press lubrication away from the bearing load zone, allowing metal to metal contact
between elements and races. If the vibration is high enough, when the primary machine fails after a
long period of service the spare machine’s bearings may fail due to brinelling damage shortly after it is
put on-line and before the primary machine is repaired. A preventative measure is to run the backup
machine periodically.

Electrical Current (EDM)

Improper electrical grounding can cause a current to flow through rolling element bearings. If arcing
occurs, small quantities of bearing material are vaporized in the small, high temperature electric arc.
This vaporization effectively removes material from the rolling element or bearing race or both. The
resulting pit acts as a stress concentrator that is capable of nucleating a fatigue crack, leading to
eventual failure. The damage caused by the electrical arc is similar to electrical discharge machining
(EDM), and takes place on motors with a variable frequency drive. On rare cases this can also be due to
buildup of static charge from the process fluids and materials.

1.5  Fatigue Failure in Rolling Element Bearings

Assuming that a bearing is properly loaded, clean, and supplied with a correct amount of clean
lubricant of the proper type; failure of a typical rolling element bearing is still inevitable because of
fatigue. Rolling element-bearing loading differs substantially from the type of loading that is found in
other engineering structures where an endurance limit can be used to guarantee freedom from fatigue
failure. Unfortunately, no such endurance limit exists for rolling element bearings, and fatigue failure of
these bearings will eventually occur.
Failure of the bearing is most likely to begin with the initiation of a subsurface fatigue micro crack in
an area of high effective, or Von Mises stress. Analysis has shown that the maximum of this stress
occurs some distance below the surface of the race or rolling element (Ref#1). See "References" on
page 36.Crack initiation is likely to be associated with some sort of discontinuity or phase change in
the metal microstructure that acts as a local stress concentrator. Alternatively, surface damage due to
debris can also produce a plastic deformation that will act as a stress concentrator. Barring surface
damage, the location of the crack will be related to the subsurface stress distribution combined with
the statistical distribution and size of material micro flaws.
As element passage produces stress cycling, the fatigue crack will begin to grow. The direction of
growth may start to spread beneath the surface, but eventually the crack will migrate to the surface of
the race or the rotating element. When the remaining material is sufficiently weakened, a spall will
break away from the surface, leaving an irregularly shaped hole, or pit. The spall itself is a metal chip
that becomes free to migrate around inside the bearing and cause additional damage. Spalls can range
in size from very small pieces to large chunks nearly as long as the width of the race.

10/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide
The initial surface failure leaves a sharp-edged hole in the race or rolling element (for this discussion,
assume that the hole is in the race and that the rolling element is a ball). When the rolling element
encounters such a hole in the load zone of the bearing, the ball will be forced into the hole. This will
produce a sudden unloading of the ball and a corresponding unloading of the race. The race will
elastically rebound slightly in the vicinity of the hole. When the ball encounters the sharp edge at the
opposite side of the hole, it will be suddenly reloaded, and the race will return to its normal, loaded
position. The sudden encounter with the edge of the spall pit causes an impact to be transmitted to
the rotor and machine structure that will generate high frequency vibration.
When the hole is fresh, the edges of the hole have relatively sharp corners, and the sudden encounter
of ball with the hole edge produces a step impulse to the entire rotor system similar to striking the
race with a hammer. The step impulse has wide frequency content and will excite the rotor system, the
housing, the casing and the transducer natural frequencies. These natural frequencies can extend to
high frequencies and produce an increased high frequency vibration that is detected with
accelerometers. The step impulse repeats when the next ball encounters the hole.

Figure 1 - 3:  Fatigue Damage Development Over 1200 Hours of Operation (Ref #6)
The very high stresses at the contact point of the ball and the edge of the hole will usually exceed the
yield strength of the material, and the edge of the hole will become peened with repeated passages
of the balls. Thus, over a longer period, the hole edge will become smoothed, and the amount of high
frequency vibration energy will decrease. The high stresses at the contact point also accelerate the
fatigue of the balls, shortening the time it will take to develop fatigue cracks.
Meanwhile, the spall is free to move around inside the bearing where it can become trapped between
a ball and a race, or between a ball and the cage. If the chip becomes trapped between the ball and
race, the local load is sharply increased, causing the outer race to deflect farther than it would under
normal loading. Plastic deformation of the race and ball is likely during this period because of the very
high local stresses. Additionally, the very high stresses will accelerate the formation of more fatigue
cracks.
Thus, the formation of a single defect will usually accelerate the formation of additional defects in the
bearing. Damage will spread from races to rolling elements to the cage in a relatively short time. As
the damage spreads and multiplies, the high frequency vibration energy will tend to increase until the
damage becomes so extensive that gross peening of the nearly totally destroyed race reduces high
frequency energy again. Thus, in the final stage of bearing failure, high frequency vibration may
decrease. For this reason, high frequency vibration, like demodulation, should not be used as a sole
bearing failure indicator.

11/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide
In the final stage of failure, gross wear of the rolling elements and the races will open the bearing
clearances. This will reduce the constraint on the motion of the rotor shaft, and, because rolling
element bearing machines are built with close internal clearances, a rotor to stator rub becomes a real
possibility. Bearing temperatures are likely to increase because of locking and sliding of the rolling
elements and the inability of the lubrication film to keep the elements separated. Noise and lower
frequency vibration will also increase.

1.6  Failure Stages

There are four main failure stages for rolling element bearings and each stage is defined by a
frequency range. These stages have been defined by Technical Associates of Charlotte, P.C. (Ref#5) and
are described below. See "References" on page 36.

Stage 1
This stage of bearing failure appears at ultrasonic frequencies ranging from approximately 20kHz
– 350kHz. This is the earliest failure stage and indicates slight defects that are not visible;
subsurface fatigue micro cracks appear as described above. Bearings should not be replaced if
the defect frequencies are shown only in this high frequency range, typically by use of
enveloping. As wear progresses, the defects are manifested in lower frequency data.

Stage 2
As bearing defects become larger, they ring the natural frequency of the bearing components
and sometimes the bearing support/machine casing. As a result, this stage occurs in the 500 –
2000 Hz frequency range. Sideband frequencies around the component/casing natural
frequencies begin to show at the lower end of the frequency range due to progressive wear.
Higher frequency information is also contained in the spectrum. Enveloping techniques are used
to detect the defect frequencies, and the acceleration waveform and spectrum may have
indications of the bearing problem.

Stage 3
During this stage, defects and wear become visible and the need for bearing replacement is
imminent. Bearing defect frequencies appear now also on the velocity spectrum. As wear
progresses, well-formed sidebands accompany the defect frequencies and harmonics.
Sidebands also continue to grow around the bearing component natural frequencies. The high
frequency content evident in stages 1 and 2 is now even higher in amplitude/energy.

Stage 4
This stage is comprised of mainly rotor-related frequencies (i.e., 1X, 2X, 3X, and 4X RPM). When
bearing defects progress to this stage, increases in rotor vibration occur. This causes the bearing
component natural frequencies and some of the bearing defect frequencies to decrease in
amplitude. However, more random broadband high frequency vibration develops, resulting in an
increasing noise floor. Just prior to total bearing failure, amplitudes in the highest frequency
region defined in stage 1 may grow excessively. If a bearing reaches this stage, damage to other
machine components (e.g., rotor, seal, coupling) is highly possible.

12/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

Figure 1 - 4:  Four Stages of Bearing Life

1.7  Bearing Defect (Fault) Frequencies

As the bearing evolves toward fatigue failure, spalling occurs on the outer race, inner race, or elements.
As rolling elements encounter spall pits (or as damaged elements rotate), the impact/response
vibration tends to repeat at characteristic frequencies, called bearing defect frequencies, often referred
as fault frequencies. Each bearing has a set of defect frequencies associated with inner race, outer race,
element, and cage defects that are determined by its particular geometry and operating speeds.
For an outer race defect (with the outer race fixed), the sequence of ball/hole impacts produces a
series of more or less constant amplitude impulse/response vibration signatures that repeat at the ball
pass frequency. For an inner race defect, the defect on the race will rotate in and out of the load zone.
When the defect is in the load zone, the impulse/response vibration amplitude will be relatively large;
when the defect is out of the load zone, the amplitude of the vibration will be relatively low or
nonexistent. Thus, the amplitude of the impulse/response vibration is modulated at the 1X vibration
frequency of the rotor. This modulation can be a clue to the location of the defect. Repeating
impulse/response vibration is used in the acceleration enveloping (demodulation) algorithm to detect
bearing defects. Modulation will also occur in the case of a stationary outer ring when the dynamic
load (i.e. unbalance) becomes large relative to the static load. In those cases where the inner ring is
stationary and the outer race rotates, modulation will also occur.

13/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide
In all cases, knowledge of the existence and extent of bearing damage is vital. Once damage has
occurred, the flaw on any element rapidly generates secondary flaws on the other elements. It is
important to detect that a bearing is failing so action can be taken in time to prevent more extensive
damage. Frequency analysis is helpful, but in most cases, precise knowledge of which part of a bearing
is failing is not as important because the entire rolling element bearing is normally replaced and not
just the damaged portion. However, when evaluating the safe remaining operating time for a REB it is
important to know which part of the bearing is failing. The general order of tolerance to a failure is
(highest to lowest): Outer ring – Inner ring – Ball – Cage. This means that with outer ring failure the
replacement is typically less urgent than with ball failure.
In a complex system like a multistage gearbox, it is also important to identify which bearing is failing.
In a system with different types of bearings, frequency analysis and high frequency detection methods
can be very helpful in identifying the problem bearing. System 1 Evolution (S1 Evo) contains
information on more than 30,000 bearings. If the needed information is not available in S1 Evo but the
bearing geometry is known, the bearing fault frequency equations on Figure 5 can be used to estimate
the defect frequencies. High frequency vibration is likely to decrease with distance, which can also help
identify which bearing is failing.
The equations in Figure 5 below have been derived using the bearing geometry and assuming there is
no slippage (sliding) of elements (Ref#1). See "References" on page 36.When using these equations,
remember that slippage may occur and modify the frequencies. Also, contact angle may change
depending on the axial load and result in a change in frequencies.
Note that these frequencies may, and often do, include sum and difference frequencies. These
frequencies will appear when the amplitude of the fundamental defect frequency is modulated by
some other rotating element - often rotor speed. Any combination of flaws can generate or be
modulated by the cage frequency, i.e. the frequency at which the elements revolve as a set.
In most applications, the outer race is fixed and does not rotate. However, in some applications, just
the outer race, or even both races rotate. We will first present the result of the typical case where the
outer race does not rotate (is fixed) in Figure 4 below. Then we will present the general case where
both races rotate (labeled free in the equations).
Finally, it is important to remember that when both races are free to rotate, they can rotate in either
direction, depending on the machine design. With these equations, use a consistent approach to
define the direction of motion. For example, use positive speeds to define rotation in a
counterclockwise (X to Y) direction and use negative speeds to define clockwise (Y to X) rotation.
The frequency equations can produce positive or negative numbers. The sign represents the direction
of the motion relative to the observer’s location given the direction convention used. The sign in the
final result does not matter for frequency analysis.

14/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide
The following symbols are used in the equations:

Symbol Description
No outer race angular speed in revolutions per minute (rpm)
Ni inner race angular speed in rpm
ր number of rotating elements
α (alpha) contact angle
D bearing pitch diameter
ɗ rolling element diameter
ʄ frequency in hertz (Hz)

A common group of factors appears in all the frequency equations. To simplify the expressions, define
ƴ (gamma) as:

Rolling element bearings have specific defect frequencies depending on the size of the bearing
components and the speed of rotation. REBs are comprised of an inner race, balls or rollers (needles,
etc.), a cage, and an outer race as shown the following figure.

Figure 1 - 5:  Rolling Element Bearing Structure

When damaged, these components produce vibration at fundamental frequencies related to the
bearing’s basic geometry. These formulas assume a rotating inner race and a fixed outer race.
 

15/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

Figure 1 - 6:  Frequencies (Hz) Produced by REB (Outer Race Fixed, No Slip)

NOTE
Earlier Bently Nevada literature may use ORBP (Outer Race Ball Pass) instead
of BPFO, and IRBP (Inner Race Ball Pass) instead of BPFI.

In all cases when identifying defect frequencies, it is very important to know the correct shaft speed.
When bearing information is not known, one clue to identifying bearing defect frequencies in the
spectrum is that they are non-integer multiplies of the shaft speed. A review of the S1 Evo bearing
database, where fault frequencies are expressed as multiples of shaft speed (see the following table), is
a good way to investigate this.

Table 1 - 1:  Defect Frequencies of Number 22320 Bearings Shown in S1 Evo Bearing

Database

Figure 1 - 7:  Defect Frequencies of Number 22320 Bearings Shown in S1 Evo Bearing

Database

16/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide
It is also worth noting that bearings with the same number but from different manufacturers may
produce different fault frequencies. A customer’s CMMS may or may not have accurate information
identifying which bearing is actually in use, so one should never confine the identification of a bearing
fault to one manufacturer’s defect frequencies.

Fundamental Train (Cage) Frequency (FTF)

The cage frequency, ʄc, is the frequency corresponding to one complete revolution of the bearing
cage relative to a fixed reference. The cage frequency is often called the fundamental train frequency
(FTF). System 1 Evo recognizes both FTFO and FTFI, cage frequencies relative to the outer and inner
rings, respectively. Very often the latter is not taken into consideration.

FTF both races free (Hz)

FTF outer race fixed (Hz)

Outer Race Element Passage Frequency (BPFO, or ORBP, or EPx)

This is the frequency at which the rolling elements pass a point on the outer race. When the outer race
is free to move, the reference point moves with it.

ORBP both races free (Hz)

ORBP outer race fixed (Hz)

Inner Race Element Passage Frequency (BPFI, or IRBP)

The BPFI is the frequency at which elements pass a point on the inner race.

BFPI both races free (Hz)

BFPI outer race fixed (Hz)

Element Spin Frequency (BSF)

The element spin frequency is the rotation speed of an individual rolling element. A defect on the
element can produce a once per turn disturbance on either the inner or outer race at the Element Spin
Frequency. It can also produce a disturbance on both the inner and outer race, producing a 2X Element
Spin Frequency.

17/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

BSF both races free (Hz)

BSF outer race fixed (Hz)

Example of defect frequencies

The following figure is an example of how defect frequencies show up in a demodulation spectrum
for a FAG 6004 bearing and a motor speed of 1970 rpm.

Figure 1 - 8:  Defect Frequencies (6004 BPFO) Shown with System 1 Evo

18/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

2.  Transducers Types and Usage

2.1  Transducer Location and Quantity
General
From an engineering principle standpoint, one would want to monitor machines with REB the same
way as critical journal bearing machines are monitored (vibration transducers at each bearing, bearing
temperature, Keyphasor sensor, online monitor). However, oftentimes it is economically difficult to
justify that level of monitoring.
The best method for monitoring REB machines is online monitoring - using either a permanently
installed continuous or scanning system. With an online system, a proper transducer suite is needed to
detect when a machine is having a problem. See "Number of Sensors" below.
In the absence of an online monitoring system, a portable data collector (PDC) is the next best
approach.

Number of Sensors
The recommended transducer suite for online monitoring of rolling element bearing machines is one
vibration transducer mounted on each bearing. If the distance between outboard and inboard bearings
is less than 1 meter (3 feet) and there is a solid metal connection between the bearings (as is the case
with overhung pumps, electric motors and with many other machines), then the number of vibration
sensors may be reduced to one per machine case. In rare cases more than one transducer per bearing
is justified.
The transducer should be mounted on or near each of the bearings in a location that provides the best
reading; there should be minimal number of interfaces between transducer and bearing. Please refer
to the following figure.

19/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

Figure 2 - 1:  Correct and Incorrect Sensor Installation Locations

For a portable data collector, the data collection points are chosen to accurately display and predict
the condition of the machine. Typically, measurements are taken in 3 directions (horizontal, vertical,
axial) for each bearing. But after a minimum of 10 surveys, the number of measurement directions per
bearing can usually be reduced to one (horizontal), with one additional axial measurement from the
bearing on either side of the coupling (see below).

Figure 2 - 2:  Recommended Initial Measurement Points with PDC and Reduced Scope
(Red)

NOTE
The above machinery measurement recommendations are guidelines only. Due to machine
construction and environmental operating conditions, the monitoring package needs to be uniquely
addressed for each machine or group of machines. In many situations these guidelines will be
adequate.

20/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

Speed Information
As mentioned before, knowing the correct shaft speed is very important when identifying bearings
fault frequencies.
For each variable speed machine monitored by an online system the speed needs to be available
either by a Keyphasor or as an OPC tag from control systems. For constant speed machines there is no
need to have online speed data; the speed needs only to be entered into the condition monitoring
system during configuration phase.
When a PDC is used with variable speed machines then the speed at the time of data collection needs
to be identified using a tachometer, stroboscope or by other means. With System 1 Evo it is also
possible afterwards to manually enter the correct speed info.

2.2  Sensor selection

An accelerometer is the recommended sensor for condition monitoring of rolling element bearings.
There are several reasons for this:

 l Accelerometers give the earliest indication of a bearing problem (see the following figure). A
lubrication problem will often lead to a bearing failure and, with a properly installed
acceleration measurement, it can be detected early. See "Stage 1" on page 12. If a lubrication
problem is corrected at this stage and before actual bearing deterioration has started, the
maintenance work and operational interruption related to a bearing failure and replacement can
be avoided. When symptoms of a bearing problem are detected with a Velomitor, Motor
Current or Temperature probe, the physical damage may have already begun and the bearing
change cannot be avoided. Refer to stages 3 and 4. See "Stage 3" on page 12.
 l Demodulation (enveloping) is done using an acceleration signal.
 l Alarm levels are based on acceleration readings. See "Alarm Levels" on page 29.
 l The acceleration signal can be integrated to velocity to provide additional diagnostics. Thus,
machinery problems like unbalance and misalignment that may lead to bearing failure can be
detected and corrected. The use of the velocity signal for protection will be dependent upon
the application and OEM recommendations.
 l The acceleration signal integrated to velocity can be used to evaluate the failure indicators in
stages 3 and 4.
 l The acceleration waveform will give additional information about which part is failing and the
severity of the problem.

21/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

Figure 2 - 3:  REB P-F Curve with Earliest Failure Detection Point for Different Sensors
There may be other machine monitoring and protection reasons that require the use of sensors other
than an accelerometer, but in those cases, failure stages 1 and 2 cannot be detected.
A brief description of different transducer types is provided below.

Accelerometer

Accelerometers measure the machine vibration in acceleration units and are mounted on the bearing
housing. Bently Nevada has several accelerometers than can be considered. All these sensors are
piezoelectric type with an integral charge amplifier. A summary of the most commonly used Bently
Nevada accelerometers for REB applications are shown in the following table.

Table 2 - 1:  Most Commonly Used Bently Nevada Accelerometers

22/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

An accelerometer is the only sensor that can be used to monitor the condition of rolling element
bearings and predict failure over all four failure stages. An accelerometer is also needed when
enveloping is required, a technique that has become an industry standard for REB monitoring. The
lower frequency content of the acceleration signal can better be analyzed when integrated to velocity.
Therefore, all instruments used with an accelerometer should support acceleration, acceleration
enveloping and velocity readings from the accelerometer sensor.
Accelerometers are relatively easy to install and involve minimal resources. However, accelerometer
performance is very sensitive to mounting location and technique, and a poor installation can very
easily degrade the performance and frequency response of the accelerometer. Signal attenuation
occurs with multiple material interfaces between a bearing and the transducer. See "Number of
Sensors" on page 19. Recommended mounting techniques will be discussed later.
Long accelerometer cable runs will degrade the high frequency signal quality. Verify the effect of field
cabling on the measurements in advance. Also, hazardous area requirements may influence field
cabling and sensor selection. See "Cables " on page 28.
When using a PDC, a single accelerometer can easily be used on multiple machines using quick
connection mounts or Mag-Base mounts. However, trending accuracy of high frequency data is
sensitive to small variations in mounting location or technique. Consistent measurements must be
taken, or trended values may be invalid and/or misleading. See "Mag-base" on page 28.

Velomitor Seismic Transducer

Velomitor piezo-velocity sensors are designed to measure absolute bearing housing, casing or
structural vibration velocities. Velomitor sensors incorporate a solid-state design and are specialized
piezoelectric accelerometers with embedded integration electronics. A summary of the most
commonly used Bently Nevada Velomitor sensors for REB applications are shown in the following
table.

Table 2 - 2:  Most Commonly Used Bently Nevada Velomitor Sensors

Velomitor sensors can be used to monitor the condition of rolling element bearings and typically
predict failure when the bearings are likely to have visible damage (stage 3). In order to get the earliest
indication of a failure it is recommended that the measurement has the highest frequency range as
possible, with peak detection used instead of rms.
The Velomitor sensor is not recommended to be used with a PDC.

23/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

Motor Current
Motor current monitoring using AnomAlert or another instrument will detect bearing problems when
the bearing failure is affecting the motor rotor / stator relationship. Thus, it is much less sensitive to
detection of early phase bearing problems than other REB vibration systems, especially if the bearing
problem is on the driven machine. It can detect bearing faults in stage 3, when the bearing fault
frequencies are detectable in the motor current spectrum.

Temperature
Thermocouples and Resistive Temperature Devices (RTD’s) are often employed for bearing
temperature monitoring. Temperature measurement cannot be relied on to independently identify
bearing faults. It may be difficult to mount the sensor close to the bearing due to the thickness of the
casing or bearing pedestal. Many factors outside the machine can also influence the temperature
readings, including ambient temperature and ambient airflow. The operator needs to be aware of all
factors influencing the temperature reading and take them into consideration when a temperature
change is noted. Nonetheless, temperature measurement should be considered if bearing protection
is required, as the temperature typically increases as the bearing enters stage 4, nearing failure. See
"REB P-F Curve with Earliest Failure Detection Point for Different Sensors" on page 22.
Thermocouples and RTD’s are much less complex in design than vibration transducers, and as such they
are less expensive and generally very reliable. Also, their small physical size allows them to be
employed when space is limited. Finally, they require almost no power to operate.
Bently Nevada does not manufacture temperature transducers. Temperature sensor specifications are
manufacturer and model specific. See www.omega.com or www.minco.com for examples.

Roller Element Bearing Activity Monitor (REBAM)

In the early 1980’s Bently Nevada developed the REBAM methodology as a method of direct
measurement of roller element bearing analysis [7]. This technique was first demonstrated by G. J.
Phillips using fiber optic sensors observing deflections in the outer race of roller element bearings [8].
See "References" on page 36. Bently Nevada developed high sensitivity eddy current sensors to
measure the micro-inch deflections in the outer race of a REB. REBAM can provide significant
improvement in sensitivity and noise reduction [9], but REBAM was not widely adopted by industry due
to the need to modify the bearing housing for installation of the REBAM probe. The REBAM probe and
associated monitoring hardware are obsolete and no longer offered by Bently Nevada. However, there
may be applications where the REBAM technology is still being used, so the following is offered for
the purpose of awareness only. Refer to References 7, 8 and 9 See "References" on page 36.

Fault Detection
REBAM probes are used to monitor the condition of rolling element bearings and predict failure only
when the bearings are likely to have visible damage. Refer to stage 3 See "Stage 3" on page 12.

Drawbacks
Since REBAM probes must view the outer ring, the bearing housing must be drilled and tapped to
accommodate the probe. Installation is time-intensive and expensive when compared to casing-
mounted transducers. Sometimes a relief hole around the probe tip must be used and if it exceeds the
maximum allowable for the bearing, it can contribute to a cyclic stress failure of the outer ring. Also,

24/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide
REBAM probes will not work if the outer race rotates or if there is a loose fit between the bearing and
its housing, such as with squeeze-film damper bearings. Neither will REBAM probes work properly with
triple ring bearings.
Predicting bearing failure after there is visible damage, as REBAM probes do, can be considered either
a drawback or an advantage depending on customer preference. Although there is less advanced
warning with REBAM than with other methods, when a bearing is indicated as defective, the defects are
generally serious enough to warrant quick replacement. Poor lubrication or excessive wear is indicated
by high frequency signals that are not likely to be picked up by displacement probes until
considerable damage has been done. Also, inner race defects and corresponding signals are
attenuated since the REBAM probe looks at the outer ring. The signal must pass from the inner race,
through a rolling element, and to the outer race.

Benefits
REBAM may have been employed when casing mounted transducers had proven not to work for the
application.
Signal-to-noise ratio (SNR) improvement was obtained using REBAM (Ref #7). See "References" on
page 36.The REBAM vibration signal is separated into two main regions: Rotor Vibration Region and
Prime Spike Region. In addition, because each REBAM probe looks directly at one bearing, there is no
confusion as to which bearing has a problem.

25/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

2.3  Accelerometer Vibration Transducer Mounting

There are three main mounting methods used:

 l Stud
 l Quick connection adapter
 l Magnet

Occasionally other methods are used, like an extension tip (stinger), adhesive, and even beeswax.
The mounting method has a significant effect on vibration signal linear response (see following figure).

Figure 2 - 4:  Transducer Mounting Effect on Sensitivity

26/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

Stud-Mount
This type of mount gives the best coupling between a transducer and the casing. It requires a drilled
and tapped hole in the machine casing into which the transducer or adapter stud is threaded (see the
following figure). This mounting gives the full frequency response range for the accelerometers and
Velomitor sensors shown in Tables 2-1 and 2-2 above.

Figure 2 - 5:  Dimensions and Tolerances for Stud Mounting

A Quick Connection Adapter

This mounting type is preferred when a PDC is used. The mounting technique requires either a drilled
and tapped hole in the machine casing into which a stud is threaded, or an adhesive to directly bond
the quick connection adapter to the machine case. The quick connection stud should be mounted on a
flat surface and is usually bonded with a quick setting epoxy like J-B Weld. This mount style has a +3dB
point at approximately 6-7 kHz when the quick connection stud is attached. Both pieces of a quick
connection adapter - the stud and base shown in the following figure - is threaded on two sides (i.e.,
two of the four quadrants are threaded). Due to cost this solution is not widely used.

Figure 2 - 6:  A Quick Connection Adapter Stud and Base, Bently Nevada Part # 46122

27/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

Mag-base
Magnetic (Mag) base mounts (see following figure) are commonly used in walk-around programs due
to the ease of installation and the variety of casing locations that can be tested with minimal mounting
preparation. Depending on the level of coupling between the magnet and the casing, the +3dB point
can be anywhere between 1 to 8 kHz. The latter requires a very smooth surface and a strong flat
magnet. Most common is a two- pole magnet, which under good conditions gives a +3dB point
around 2 kHz. When a magnetic base mount is used for routine data collection, the measurement point
should be marked in some manner to consistently locate the sensor during each measurement. Paint
markings around a mag-base mounting location, or a small punch mark, will promote measurement
repeatability.

Figure 2 - 7:  Sensor on Magnetic Mount

Cables
Sensor cables between an accelerometer and the charge amplifier are susceptible to creating their
own acceleration signal due to the triboelectric effect if they vibrate. When installing an accelerometer
that requires an external charge amplifier, route the accelerometer cable away from the moving
components of the machine and avoid small bending radii in order to extend the cable life.
Long cable runs will reduce the overall system’s capability to measure high frequency readings. Use the
Bently Nevada field wiring calculation tool to define the system maximum frequency for a given field
cable length and capacitance.

28/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

3.  Alarm Levels
Since the REB is always an integral part of the machine in which it is used, there is no industry standard
for alarm levels associated with a particular REB. The machine OEM’s recommendations should always
be followed for Alert and Danger alarm levels.
Bently Nevada does not make recommendations for machinery protection Alert and Danger levels, but
can work with end users and OEMs to acquire and evaluate engineering data so the end user can set
preliminary alarm levels
In the absence of OEM recommendations or engineering data on a machine, some of the factors that
influence vibration levels are:

 l Machine speeds
 l Bearing types
 l Load on machine and bearings
 l Environment temperature
 l Signal path from bearing to sensor
 l Sensor connection method
 l Failing part in the bearing

ISO 13373-3
A draft standard currently under evaluation is “ISO 13373-3 Condition monitoring and diagnostics of
machines - Vibration condition monitoring. Part 3: Guidelines for vibration diagnosis”. Annex D titled
“Diagnosis of rolling element bearings” contains proposed alarm levels for acceleration measurements
from 10 Hz – 10 kHz, both in pk and rms values. The relationship of these values (very much like Crest –
factor, see 4.2) is the basis of the alarm levels. These values are not recommended to be used for
protection. Note that in order to get reliable measurements up to 10 kHz the accelerometer must be
stud mounted. This standard may be accessed at http://edms.pw.ge.com/trslibrary/

Crest factor
Crest factor is defined as the ratio of zero-to-peak amplitude of a waveform to the rms value of the
same waveform (see following formula ). It can be described as a measure of the “spikiness” of the
waveform. The rms value of a purely sinusoidal waveform is A/√2, so the crest factor would be A/
(A/√2), which equals to 1.414. In general, a high crest factor means the waveform contains impact type
events. In the context of rolling element bearings, a high crest factor is an indication of a bearing
defect. In a gearbox application, it may be an indication of a cracked tooth.
The crest factor measurement has been used to detect impact symptoms in vibration signals for many
years. It has been used in Bently Nevada ADAPT.Wind and in S1 Evo for any waveform signal. For REB
purposes it is calculated from an acceleration waveform with a frequency setting of Fmax ~50 X using
formula:

29/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide
Based on field experience, and in the absence of OEM recommendations or other engineering values, a
crest factor level of 4.0 can be considered as an initial alarm level. It is not recommended to use crest
factor for protection, as it is mainly targeted for early warning purposes and typically will decrease as a
bearing failure moves into its final stages.
Crest factor can also be calculated from enveloping (demodulation) waveforms, but since the signal
processing will remove lower frequency components, the crest factor will be much higher than in an
acceleration waveform, and a higher alarm levels should be applied.

Enveloping (Demodulation)
Enveloping is an early indication method and as such should not be used for protection. Bently Nevada
does not promote any Enveloping alarm levels. Other vendors publish Enveloping alarm levels, but due
to different filter settings, peak detection techniques and other signal processing related issues those
cannot be directly applied.
In general, Enveloping alarm levels are normalized by shaft speed and diameter. This means that the
larger the bearing diameter and the higher the speed, the higher the alarm level.
The enveloped peak-value usually gives a very early indication of bearing problems. On the other
hand, the enveloped rms-value, sometimes referred as carpet value, has proven to be a good indicator
of a bearing lubrication problem. If the bearing lubrication was perfect, there would be no metal-to-
metal contact within a bearing and the enveloped rms-value would be close to zero. Because there is
always some level of metal-to-metal contact, all bearings will have a non-zero enveloped rms-value.
An increase of the enveloped rms-value for a particular bearing might indicate under-lubrication,
improper lubrication, or even increased bearing loading due to misalignment of the rotor or an
unbalance. For both peak and rms values to be useful during trending, a baseline must be taken when
the bearing is known to operate correctly.
For a better understanding of the enveloping process, refer to Bently Nevada Basic Vibration Analysis
training, Document 176105, Signal Processing.

30/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

4.  Vibration Settings for Bently Nevada Monitors in

REB Applications
This section gives guidelines for optimizing hardware settings to collect the best possible data for
REB. It must be noted that other monitoring and protection priorities may prevent these settings from
being implemented in full.

4.1  3500 Monitor

If the preferred sensor for REB, an accelerometer, is installed, it is recommended that a channel pair is
reserved for a single input by enabling “Take Input from Channel 1 Transducer” in the Channel 2 or 4
Transducer Selection option. One channel will measure integrated Velocity, thus monitoring rotor
related issues and REB failures at stage 3 and 4. The other channel measures the acceleration pk value,
thus giving an earlier warning of bearing failure.
If a Velomitor is installed, then only a single channel is needed. Fmax should be set to 50X and pk –
units should be used.

4.2  3701 Monitor

The 3701 monitor processes all the data needed for good REB condition monitoring from a single
accelerometer: velocity, acceleration and acceleration enveloping. All measurements derived from
waveforms and spectrums can be individually configured using suitable HP and LP filters and spectral
lines. Recommended settings are:

 l Velocity spectrum with Fmax 50X

 l Acceleration waveform Fmax 50X (primary)
 l Acceleration waveform Fmax 10 kHz (secondary) (use a stud mounted accelerometer)
 l Enveloping spectrum

Note that 3701 processing power is limited, so not necessarily all recommended measurements can be
assigned to all channels. In that case, a secondary acceleration waveform is the first one that should be
removed.

4.3  2300 Monitor

The 2300 monitor provides all the data from a single accelerometer that is needed for good REB
condition monitoring: velocity and acceleration trended values, and acceleration and demodulated
(enveloping) waveforms.
The 2300 monitor comes with five (5) predefined static variables:

 l Acceleration pk
 l Acceleration rms
 l Acceleration derived pk (rms x 1.41)

31/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide
 l Velocity pk
 l Velocity rms

These five (5) predefined variables all have the same HP and LP filters. For bearing monitoring
purposes the LP filter should be set to 10 kHz.
The user can also define two (2) additional independent bandpass variables for each input channel with
independently configurable HP and LP filter settings. Each additional variable can be in acceleration or
velocity units.
Recommended settings for dynamic data are:

 l Velocity waveform with Fmax 50X

 l Acceleration waveform Fmax 50X
 l Demodulation waveform

Note that the 2300 monitor does not calculate any demodulation-based static variables: all these
variables need to be calculated in S1 Evo, where the alarming also takes place.

4.4  1900/65A Monitor

The 1900/65A monitor allows three (3) independent static variables to be defined for each
accelerometer sensor. The recommended settings for these variables are:

Table 4 - 1:  1900/65A Settings for REB

If only one variable can be specified for REB monitoring, then either #2 or #3 should be selected
depending user preference.

4.5  VbOnlinePro Monitor

The VbOnlinePro monitor provides all the data from a single accelerometer that is needed for good
REB condition monitoring: velocity, acceleration and acceleration enveloping. All measurements
derived from waveforms and spectrums can be individually configured using suitable HP and LP filters
and spectral lines. Recommended settings are:

 l Velocity spectrum with Fmax 50X

 l Acceleration waveform Fmax 50X
 l Enveloping waveform

32/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide
If the Fmax setting is above 10 kHz, the performance of VbOnlinePro monitor will be affected (see
product user guide, document 115M4367 for details).

4.6  Trendmaster Pro Dynamic Scanning Module (DSM)

TIM-line
The recommended transducer is the Bently Nevada 200157 accelerometer, and the corresponding
TIM-module part number is 200200-06-06. This combination provides the ability to monitor both
velocity and acceleration enveloping data. These settings are recommended:

 l Transducer Tab: pk detection. This setting enables both rms and pk static values.
 l Waveforms Tab: Enable all return waveforms, both Velocity waveforms and Acceleration
Enveloping. Set Frequency Span on waveform 1 to 50X and on waveform 2 to at least 5 kHz. The
enveloping filter is fixed, and the frequency span should be high enough to cover 3 x BPFI.
 l Variables Tab: Enable all that can be used; some require the Keyphasor sensor to become active.
 l Filters Tab: Enable all filtered variables and set the frequency span according to the machine
monitoring needs.

4.7  RangerPro (Wireless)

The configuration of RangerPro sensors is done with RangerPro software. The recommended settings
are:

 l Static values:
 o Acceleration HP 5 Hz, LP 5 kHz on X and Y directions, 10 kHz on Z direction.
 o Velocity Hp 5 Hz, LP 1000 Hz on machine RPM < 1800, 2000 Hz on machines with RPM >
1800 rpm.
 o Peak Demod enabled
 o Temperature enabled.
 o Pk and RMS selection based on detection needs.

 l Dynamic waveform settings:

 o Acceleration X and Y directions 5 kHz, 10 kHz on Z direction.
 o PeakDemod waveform enabled
 o Waveform point count selected based on resolution and data transfer speed requirements.

Note that System 1 Evo allows additional spectral band and waveform static value calculations from
the dynamic data.

33/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

4.8   SCOUT Portable Data Collector (PDC)

The SCOUT PDC provides the user all required REB condition monitoring data from a single
accelerometer: velocity, acceleration and acceleration enveloping (demodulation). This data can be
collected in series, or in parallel using 6Pack functionality to reduce collection time. When System 1
Evo Quick Configuration is used, all the recommended settings are implemented automatically.
Alternatively, all dynamic data can be configured individually using suitable HP and LP filters and
spectral lines to provide all desired spectral and waveform variables. Recommended settings are:

Table 4 - 2:  System 1 Evo / SCOUT Settings for REB

Considerations for specific machine types:

 l If the machine is slow and big (as in food plants, pulp & paper, mining, etc.), then the
demodulation filter should be lower because the local natural frequencies are lower. Use a 0.5-
10 kHz or even a 0.25-10 kHz enveloping filter.
 l If the machine has a gearbox it is desirable to see the effect of the gear mesh frequency in the
enveloped spectrum. In this case, the minimum frequency of the demodulation filter should be
increased (especially for small machines).

Data collection interval

An important part of establishing a PDC route is defining how often the data is taken. The more often
the data is collected, the more likely an early bearing failure detection will be made. But as a
consequence, the cost of the condition monitoring will be higher due to more frequent data
collection and analysis. There are no industry standards to guide this process. The route creation and
timing depend on the machine operating hours, running speed, load, process and environmental
conditions, and so on. The following general guidelines may be useful in establishing the initial PDC
route:

 l A Four (4) week interval is generally recommended for the initial period
 l If the interval is longer than eight (8) weeks, advanced bearing failure predictions may no longer
be feasible
 l If the interval ends up being less than two (2) weeks, an online system may be justified

Spared machines should also be monitored. Sometimes this requires that machines are started just for
data collection purposes. In that case the machine needs to run long enough so that normal operating
conditions and temperatures are reached before data collection is done.

34/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

5.  Monitoring Slow Speed Machines

In slow shaft speed situations (below 100 rpm), monitoring rolling element bearings becomes
challenging. Low energy levels and a relatively long time between bearing failure-based impacts will
drive two changes: Velocity data becomes less important, and acceleration enveloping becomes more
important. Although the acceleration waveform is still important, it must be sufficiently long and the
sampling frequency high enough to assure that the impacting is not filtered away.
The slow speed also effects how the analysis is performed: Less value is gained from the spectral data,
and more emphasis is placed on waveforms and static value trends. When dealing with very slow
speed machines (10 rpm and below), it is likely that acceleration enveloping waveform and trending of
enveloping pk and rms values are the most important tools.

35/36 164934 Rev. C

Rolling Element Bearing Methodology
Application Guide

6.  Conclusion
This document provides key information regarding rolling element bearings - their construction and
failure modes, and the methods, equipment, and settings used to detect or avoid those failures in
typical industrial applications. Although much of it applies universally to all REB applications, there are
some unique applications (aeroderivative gas turbines, wind turbine generators, other) that are beyond
the scope of detailed treatment in this document. In those cases, or for clarification or further
information regarding anything herein, please contact a Bently Nevada Application & Solution
Architect (ASA) or Machinery Diagnostic Services (MDS) engineer.

6.1  References
1. Harris, Tedric A., “Rolling Bearing Analysis”, 4th Ed., published by Wiley-Interscience, December 15,
2000. This is considered by many to be the bible of rolling element bearings.
2. Loewenthal, S., Moyer, D, and Needelman, W., “Effects of Ultra-Clean and Centrifugal Filtration on
Rolling Element Bearing Life,” Journal of Lubrication Technology, Vol 104, page 283, July 1982.
3. Juvinall, Robert C., “Fundamentals of Machine Component Design”, John Wiley and Sons, 1983.
4. API Standard 670, 5th Ed., November 2014, pp. 108 -112. Published by the American Petroleum
Institute, Washington, D.C. http://www.api.org
5. Berry, James E., “High-Frequency Enveloping and Demodulation Techniques”, Technical Associates Of
Charlotte, PC, Charlotte, NC
6. FAG Publ. No. WL82102/2EA, “Rolling Bearing Damage Recognition of damage and bearing
inspection”
7. Performance of REBAM during ball bearing failures, Part 1 September 1992, Part 2 December 1992.
8. Philips, G. J., “The Fiber Optic Bearing Monitor”, Instrumentation in the Aerospace Industry Vol 28,
Advances in in Test Measurement-Vol 19 Part One, 1982 pp. 379-395.
9. Vibration Monitoring of Rotating Systems, Engineering, 2010, 2, 46-54 doi:10.4236/eng.2010.21006
Published Online January 2010 (http://www.scirp.org/journal/eng/).
http://file.scirp.org/pdf/Engineering20100100006_78410574.pdf

36/36 164934 Rev. C